Thursday , July 18 2019
Home / brazil / Embrapa develops technology to keep track of a herd with a drone

Embrapa develops technology to keep track of a herd with a drone



drone from Embrapa
Statue: Silvia Z. Borges / Embrapa

The activity to monitor livestock numbers is critical in managing livestock farms. However, in the large estates which absorb a very wide, widespread cattle ranching in Brazil, this calculation needs advanced technologies and processes. But the co-ordinator of the Brazilian Agricultural Research Group (Embrapa) carried out a survey for detection and counting using an unmanned air vehicle (Vant), which is also carried out as a console.

The project, led by researcher Jayme Garcia Arnal Barbedo of Embrapa Informática Agropecuária (SP), was approved for funding from the Trust for Research Support at São Paulo Stadium (Mapesp).

“Air examinations seem to be a solution,” said Barbedo. Some initiatives are already being implemented on farms, but there are still a number of disadvantages such as high operating costs, high noise levels and accident risk. So the researchers want approaches to processing photos and learning tools, along with software development and drone use, seeking more suitable solutions for animal grazing in the grazing.

Experiments with 360 degree cameras

Among the goals is the development of a methodology for using the hard drives using drones equipped with cameras with visible spectacle and the creation of an algorithm to identify the cattle in the images captured and counted automatically.

Two different methods will be explored: the first with routine camera use and the other with a 360 degree. The main difference between the two approaches is that the first one is several times to cover large areas, and the latter may be able to see much larger sections with one aircraft, although it is being solved, from which the animals are further away from the camera, t according to the researcher. t

Soon, doses estimate the health and weight of the flock

The project aims to advance in the field of computing science, preparing new algorithms based on image-searching and for its management, with an innovative way to count cattle. "This approach can also be used in the future to work on other aspects of cattle analysis such as animal health, body measures, amongst others," said Barbedo.

Embrapa Informática Agropecuária is mainly responsible for the collection and collection of images in the field and for their creation and promotion by Campinas State University of Engineering (FEEC / Unicamp) , algorithms based on learning principles. Finding and counting animals.

The trials will be carried out with the help of animal management experts at the farms of Canchim, from the Sudeste Embrapa Pecuária Sudeste. The researcher Andrea Roberto Bueno Ribeiro, from the Unidas Metropolitanas Faculty (FMU), will also support animal management studies and animal behavior surveys associated with cattle movement changes in grazing.

Counting cattle without gathering animals

It is vital that you know the number of animals so that the farmer can properly manage the farm, manage internal control, and measure its results. In the process, the differences in the herd are sustainable: birth, death, sale and even thieves. All of this is revealed during planning and profitability for the activity.

For counting, the cattle ranger needs to gather the animals in the grazings or take them to the crown. This process is complex and needs a lot of effort. Problems are equivalent to the size of a farm and the herd.

Eradication is always a concern and can affect productivity. Along the way, there are opportunities for accidents. In addition, the cattle spend more time walking and stopping, restoring and resting, causing economic loss. The cheese reduces weight and affects immunity, has a bigger impact on disease.

“The idea of ​​this project is to try to do this without putting animals together, reducing pressure. However, there are problems. It is necessary to train the device to see the stag and count it in turn. The first problem is that it is moving. The second is that the dose must recognize the ox down there and know it's an animal, not a stone or even if you have two animals together, you need to be able to understand it. counting two, not just one. The machine has to learn how to analyze the image, "explains the researcher Patrícia Menezes Santos de Southeast Livestock Embrapa.

To capture the images, one must understand the behavior of cattle. In this way, the aircraft plan and the strategies used to capture them will include the introduction of animals, types of farming, production model model, amongst other things in terms of creation and production. T property. The interaction between the experts of the two research institutions in Embrapa will enhance the best way to teach the instrument in the evaluations.

At the Canchim farm, a physical base for testing and catching images, there are several models of production, including animals in integrated systems, broad grazing, intensive fields and others. According to Embrapa, this difference will help in the device settings. “The doses need lots of images with real information about what happens on the farm. Everything can be a barrier when assessing images. So we will help as experts in animal production, ”explained Patricia.

The development of technological solutions can center on the implementation of intelligent and automated systems for the grasslands to contribute to future livestock production, because the process of decision-making of the production systems depends on greater analysis and interpretation of data and information, t According to Patrícia. “Connecting climate information, soil, stock and grazing creates opportunities for more efficient use of natural resources and fertility factors,” said the researcher.

A new call to action

Decisions about buying and selling animals, fertilizers and additional nutrients, as well as other regulatory practices may be guided by this information, mitigating the risk associated with activity and increasing the sustainability of animal production systems in grassland. With the progress of the Internet of Things (IoT) area, highlighted by the link between various devices that automatically collect and disseminate data, there is an expectation that there will be more established digital technology based stock in place. 'open.'

Support

The project is valid for two years, starting in February 2019 and funded around R $ 175,000, which will be invested in building doors and photographic processing equipment.

Have you seen it? Lightning has killed 21 animals in Mato Grosso do Sul


Source link